Evaluation of the Sensititre System for Identification of Enterobacteriaceae

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, Apr. 983, p /83/ $0.00/0 Copyright 983, American Society for Microbiology Vol. 7, No. 4 Evaluation of the Sensititre System for Identification of Enterobacteriaceae JOSEPH L. STANECK,'* JEAN VINCELETTE, FRANCOIS LAMOTHE AND ELIZABETH A. POLK3t Department of Pathology and Laboratory Medicine, University of Cincinnati College of Medicine, Cincinnati, Ohio 4567; Microbiologie, Hopital Saint-Luc, Montreal, Quebec, Canada HX 354; and Clinical Microbiology, Tampa General Hospital, Tampa, Florida Received 9 September 98/Accepted 3 January 983 The Sensititre identification system (Seward Laboratory/GIBCO Laboratories) consists of a microplate containing a pattern of 4 biochemicals repeated four times together with an automatic inoculation device and a microcomputer-assisted data interpretation component. A total of,45 isolates of Enterobacteriaceae plus 6 isolates of other glucose-fermenting gram-negative bacilli -were tested in three hospital laboratories in parallel with API 0E (Analytab Products). Discrepancies were resolved by conventional biochemical testing. Sensititre yielded correct identifications at the species level with 94.6% of the isolates and at the genus level with an additional.9%. API 0E yielded correct species identification with 9.% and genus only identification with an additional 6.7% of the isolates. For the routine identification of clinical Enterobacteriaceae isolates, the Sensititre system compares favorably with API 0E and offers clinical laboratories the economy of a microtiter plate system as well as the benefit of a microcomputer capable of other microbiological and data management applications. A significant portion of a routine clinical microbiology laboratory's workload consists of the identification of aerobic gram-negative bacilli belonging to the family Enterobacteriaceae. Since specific identification within this group often has clinical and epidemiological implication, accurate and reproducible taxonomic classification of clinical isolates belonging to the Enterobacteriaceae is mandatory. Within the past 0 years, the value of commercially prepared, computer-assisted, multitest systems has been well established in terms of quality of results, ease of use, and economy. The microtiter plate, which has been well received by clinical laboratories as a convenient vehicle for antimicrobial susceptibility testing (7) and a variety of serological procedures, also lends itself to the miniature multitest approach to bacterial identification. Several authors have commented favorably on the use of a commercial microtiter system with frozen reagents for the identification of Enterobacteriaceae (, 4). The present study assesses the performance of a new microtiter plate bacterial identification system utilizing dried test substrates (Sensititre; manufactured by Seward Laboratory, London, England, and distributed in the United States and Canada t Present address: GIBCO Laboratories, Lawrence, MA by GIBCO Laboratories, Lawrence, Mass.) in comparison with that of a commonly used multitest identification kit, the API 0E (Analytab Products, Plainview, N.Y.) (6). This study involved the identification of Enterobacteriaceae and was carried out in three independent clinical microbiology laboratories: University of Cincinnati Hospital, Cincinnati, Ohio; Saint-Luc's Hospital, Montreal, Quebec, Canada; and Tampa General Hospital, Tampa, Fla. MATERIALS AND METHODS Clinical isolates. Gram-negative bacilli resembling Enterobacteriaceae by virtue of colonial morphology on sheep blood and MacConkey agars were studied. With the exception of Aeromonas spp. and Pleisiomonas spp., all isolates were oxidase negative. In the majority of instances, identical colonies growing from clinical specimens plated day earlier on a primary plating medium were used as inoculum for both the API 0E and Sensititre identification systems. In those instances when well-isolated colonies were unavailable, reisolation on appropriate solid medium was carried out before proceeding with an identification. Consecutive clinical isolates were entered into the evaluation throughout the study period; however, to avoid a disproportionate number of Escherichia coli isolates, each study center, after recognizing approximately 00 E. coli isolates, excluded from the study additional isolates which gave characteristic E. coli morphology on MacConkey agar and were oxidase negative and spot indole positive. Downloaded from on December, 08 by guest 647

2 648 STANECK ET AL. Stock culture isolates. Each of the three study centers contributed less commonly encountered clinical isolates to the evaluation from their respective stock culture collections. Each isolate was given a numerical code by a laboratory member other than those involved in the identification study so that each stock isolate was run blindly. Each isolate was subcultured daily onto sheep blood agar from the stock holding medium for consecutive days before inoculation ihto the identification systems. Less than 5% of the study was conducted with isolates taken from stock cultures. Interlaboratory reproducibility. An additional 0 stock culture isolates were numerically coded by an individual at the University of Cincinnati not participating in the identification study. Triplicates of each isolate were prepared and distributed to the three study centers as unknowns. Each of the coded isolates was then entered into the study protocol at each of the three centers. Tabulation of responses was made later, noting both individual biochemical reactions and taxonomic classifications. API 0E system. The API 0E was used according to the manufacturer's instructions. All test strips used throughout the study were of the same lot number. A profile recognition system index (issued December 979) was utilized for taxonomic designation. For purposes of this study, identifications were considered valid only if the quality of identification given by the profile index was "excellent," "very good," or "acceptable" without additional tests. Identification at the genus level was considered valid if the report of "good likelihood but low selectivity" (GLBLS) was accompanied by a first and second choice having the same genus but different species. Otherwise, GLBLS was, in those instances where both the genus and species of the first and second choice differed, not considered a valid identification and was placed in the functional category "no ID." No attempt was made to clarify interpretations by using the telephone computer service from Analytab Products. Sensititre identification system. The Sensititre identification system was used according to the package J. CLIN. MICROBIOL. insert instructions. Descriptions are given below regarding the system components, the method of use, and the computer-assisted derivation of identification. (i) Components. The Sensititre identification plate consists of a 96-well plastic microtiter plate packaged in a moisture-proof aluminum foil pouch, containing 4 dried biochemical test substrates arranged in three vertical columns repeated four times across the plate. The first column of eight wells contained the nitrate, oxidase, indole, Voges-Proskauer, tryptophan deaminase, gelatin liquification, adonitol, and rhamnose tests. The second column of eight wells contained the glucose, o-nitrophenyl-4-d-galactopyranoside, urease, citrate, H,S, inositol, arabinose, and sorbitol tests. The third column of eight wells contained the decarboxylase broth base, lysine decarboxylase, ornithine decarboxylase, malonate, esculin, mannitol, maltose, and sucrose tests. The top row of substrates in each quadrant (nitrate, glucose, decarboxylase broth base), yielding positive, positive, and negative results, respectively, is confirmatory for members of the family Enterobacteriaceae (organisms producing other patterns were excluded from this study). The remaining seven rows are scored to yield a seven-digit biocode which contains the information necessary for identification. Reagents are supplied in dropper bottles by the manufacturer for those tests requiring the addition of reagents after incubation. Accessory materials include mineral oil, adhesive plate covers, and plastic templates for oil overlay and reagent addition. All Sensititre plates used throughout the study were from the same lot. Inoculation and hydration of the test substrates were accomplished by using the Sensititre automatic inoculator, a device that delivers 00,ul of water containing the inoculum into each of the 4 biochemical wells required for an identification. The inoculum is delivered directly from a glass tube (6 by 0 mm) with a standard threaded lip onto which is fitted an individual disposable plastic dosing head that attaches to the inoculator. Each isolate is therefore inoculated from a self-contained, aerosol-free disposable dosing Downloaded from on December, 08 by guest FIG.. Sensititre automatic inoculator.

3 VOL. 7, 983 SENSITITRE BACTERIAL IDENTIFICATION SYSTEM 649 FIG.. unit (Fig. ). The inoculator transports the microtiter plate under the dosing head such that each test well receives the appropriate inoculum volume. This sequence is repeated in each of the four test patterns, enabling the testing of four organisms per plate. The Sensititre data interpretation system consists of a plate viewer, a microcomputer (Apple II), visual display unit, two floppy disk drives, and a printer (Fig. ). Patient demographics and isolate reference number may be entered by keyboard before reading the reactions of a particular isolate. The plate viewer is designed to allow the operator to read the plate from the bottom via a mirror. A series of indicator lights illuminates in sequence the wells to be read, the results of which are entered by means of the computer keyboard. The visual display unit displays the test matrix for each identification, the information necessary to read each reaction, each test result upon keyboard entry, and the calculation of the seven-digit octal number based on the final tests of the pattern. This number is then utilized by the identification program to give a taxon designation together with a statement of identification quality. The patient demographics, the entire test result record and subsequent identification analysis, including mathematical parameters and test results against each taxon choice, are displayed on the visual display unit and can be generated in hard copy via the computer-driven printer. Sensititre data interpretation system. (ii) Method of use. Two or three identical colonies, 8 to 4 h old, were picked from a sheep blood or MacConkey plate with a wooden applicator stick and emulsified in 7 ml of sterile demineralized water to yield a cell concentration of approximately 07 CFU/ml. Delivery of inoculum to the microtiter plate by means of the automatic inoculator device occurred within 30 min of inoculum preparation. Mineral oil overlay was made to appropriate wells, transparent adhesive seals were applied, and perforations were made over all citrate and malonate wells. Four isolates could be inoculated per plate. Incubation of plates was carried out at 35 C under ambient atmosphere for 8 to 4 h. Before the plates were read, appropriate reagents were added according to the manufacturer's instructions. The oxidase reaction was read immediately, whereas the remaining reactions were read after 0 to 5 min to allow for the development of positive Voges- Proskauer reactions. To facilitate uniform reading of reactions between the three study centers, plates were read in the following manner at each center. All reactions were read on the plate viewer by using a fluorescent desk light situated above the viewer, diffused by a /8 in. (3. mm) thick white translucent Plexiglas sheet (4 by 6 in. [0. by 5. cm]) suspended approximately 6 in. (5. cm) above the plate viewer platform. This illumination provided bright reaction colors against an even background. For carbohydrate fermentation reactions, only yellow was considered positive, whereas blue or any shade of green was considered negative. The decarboxylase reactions were read by comparison with the decarboxylase base broth negative control well. To be called positive, the decarboxylase test well had to be purple and those wells which were not purple, even if darker than the control well, were called negative. (Uii) Computer-assisted derivation of identification. The computer-assisted derivation of identification utilized a seven-digit "biocode" generated by scoring the reactions of rows through 7 of the plate for each organism tested. The first, second, and third reaction wells of a given row were valued at four, two, and one, respectively, if positive, and zero if negative. The sum of these three values represented one number of the code. The biocode is referred by the computer to a data base compiled from the testing of over 40 taxa of gram-negative organisms. The identification is obtained by considering absolute likelihood (8) and modal distance probability values (R. K. A. Feltham and M. Stevens, in C. S. Gutteridge, ed., New Methods for Detection and Characterization of Microorganisms, in press). Likelihood values dictate the order of taxon choices, whereas the quality of the identification statement (excellent, good, etc.) is derived from the modal distance values. For purposes of this study, taxonomic designations were considered valid if the quality of Sensititre identi- Downloaded from on December, 08 by guest

4 650 STANECK ET AL. fication was excellent, good, acceptable, or "rare biotype." Occasionally, when the mathematical model indicated that an isolate was determined to be located at the periphery of two overlapping modal distance clusters, a listing of the choices would appear with the comment "low selectivity." In these cases, the first choice was not considered a valid identification for purposes of this study. Occasionally, a low selectivity comment would accompany a listing in which choices and were of the same genus but separation at the species level could not be made mathematically without further testing. Such instances were accepted as valid identification at the genus level only. Conventional identification of bacterial isolates. Isolates for which the initial API 0E and Sensititre identification responses differed, either at the genus or species level, were subject to repeat testing on both systems. Persistent discrepancies were resolved by identification of the isolates with conventional tubed biochemical tests. Test results as reported by Edwards and Ewing (3) and updated by Brenner et al. () were used to determine identification. Nomenclature as described by Skerman et al. (5) was utilized throughout the study except for Salmonella arizonae, which is referred to in this report as Arizona hinshawii. The following biochemical tests or test media were routinely used: triple sugar iron agar, decarboxylation of lysine or ornithine, motility, gas production, HS production, utilization of citrate, urease activity, production of indole, deamination of phenylalanine, and cytochrome oxidase activity. Additional biochemical tests were used when necessary. Incubation times appropriate to each test were utilized. Serological grouping with antisera directed against somatic antigens was performed when indicated for suspected isolates of Salmonella and Shigella. Study protocol. Isolates were examined simultaneously on API 0E and Sensititre identification systems. Those isolates which yielded identical species identification responses from the two systems were subject to no further identification procedures. The initial agreement response was considered to be the correct taxon for that organism. Isolates not in agreement between the two systems, either at the genus or species level, or isolates yielding a "genus only" response on one or the other identification system were repeated on the two systems, and if disagreement persisted, they were submitted to conventional biochemical testing. In these instances, the identification response made by conventional biochemical testing was accepted as the correct taxonomic classification. Any time the genus or species response of an isolate was identical on both the API 0E and the Sensititre, or when the response of either API 0E or Sensititre was identical to that given by conventional biochemical testing, the response was considered for tabulation purposes as "complete agreement." If the response of either API 0E or Sensititre indicated the correct genus but either the comment "species" or low selectivity (GLBLS on API 0E) at the species level, the response was scored as a "partial agreement." A no ID category included those instances when the code number generated did not appear in either the API 0E profile recognition index or the Sensititre computer data base or when the response by either system was low selectivity at the genus level. In those instances when a test system gave a valid response that differed J. CLIN. MICROBIOL. from that of the conventional biochemical testing at the genus or species level, the response was tabulated as an "incorrect ID." RESULTS A total of,4 gram-negative bacilli representing 35 different taxonomic groups were examined in the study. There were,53 isolates which yielded identical taxonomic responses on API 0E and Sensititre, including 47 isolates whose initial disagreement was resolved upon repeat testing. The remaining 68 persistently discrepant isolates were studied further with conventional biochemicals. The responses of each identification system were tabulated, and the distribution of all identification responses is summarized in Table. Correct identification at the species level was made for 9.% of the isolates by API 0E, whereas correct identification was made at the species level by Sensititre for 94.6% of the isolates. Correct identification at the genus level, including instances where only the genus was given as the response (partial agreement columns, Table ) and instances where the response included the correct genus but incorrect species designation (Table ), was made for an additional 6.7 and.9% of the isolates by API 0E and Sensititre, respectively. The nature of the individual incorrect responses by either API 0E or Sensititre is presented in Table. Results of conventional biochemical testing were utilized to designate the correct taxonomic classification. The API 0E gave 4 such incorrect responses involving six taxa, including six instances in which Serratia liquefaciens was called Serratia marcescens. There were eight instances involving seven different taxa in which Sensititre yielded an incorrect identification. Twenty coded stock isolates consisting of a variety of Enterobacteriaceae were submitted to each study center to determine interlaboratory reproducibility. The number of instances where one laboratory read a specific biochemical reaction for a particular isolate differently from the other two laboratories was only 8 and 6 for API 0E and Sensititre, respectively. The total number of individual reactions contributing to the identification read on each system for the 0 isolates in the three laboratories was,60. In no cases did this variation in individual reaction reading influence final identification responses on either system. Nor did any one test reaction on either system appear to be significantly more subject to variation in reading between laboratories than any other reaction. Twenty identical identification responses were noted among the three laboratories within each test system. However, the two systems differed slightly in their response to a Citrobacter diversius isolate. Downloaded from on December, 08 by guest

5 VOL. 7, 983 SENSITITRE BACTERIAL IDENTIFICATION SYSTEM 65 TABLE. Performance of the API 0E and Sensititre systems in the identification of Enterobacteriaceae Total API 0E Sensititre Taxon' no. of Complete Partial No Incorrect Complete Partial No Incorrect isolates agreementb agreement' IDd ID agreement agreement ID ID Aeromonas hydrophila Arizona hinshawii Citrobacter diversus Citrobacter freundii Edwardsiella tarda Enteric group 7 Enterobacter aerogenes Enterobacter agglomerans Enterobacter cloacae Enterobacter gergoviae Enterobacter sakazakii Escherichia coli Hafnia alvei Klebsiella ozaenae Klebsiella oxytoca Klebsiella pneumoniae Klebsiella rhinoscleromatis Morganella morganii Plesiomonas shigelloides Proteus mirabilis Proteus vulgaris Providencia alcalifaciens Providencia rettgeri Providencia stuartii Salmonella choleraesuis Salmonella enteritidis Serratia marcescens Serratia liquefaciens Serratia rubidaea Shigella boydii Shigella dysenteriae Shigella flexneri Shigella sonnei Yersinia entercolitica Yersinia pseudotuberculosis S a Taxonomic classification determined by initial agreement of API 0E and Sensititre or, in those instances of initial disagreement, by evaluation with conventional biochemicals. b Agreement at both genus and species level. c Agreement at genus level; commercial system responds species or low selectivity at the species level. d Code number not in data base or low selectivity response for which first and second choices differ at the genus level. Whereas Sensititre reported C. diversus, API 0E reported Citrobacter species. In addition, both API and Sensititre reported identifications at the genus level only for a Shigella boydii and a Shigella flexneri isolate, but again each system gave identical responses in each of the three laboratories for these two isolates. DISCUSSION The results of this study indicate that the Sensititre system compares favorably to a wellaccepted commercial multitest system, the API 0E, for routine identification of Enterobacteriaceae. Correct identification to the species level was given by Sensititre for approximately 95% of the isolates tested, whereas API 0E yielded a correct species identification in 9% of the isolates. The difference between the two systems at this level of performance can be largely attributed to the ability of Sensititre to yield acceptable species identification for Citrobacter diversus and Salmonella enteritidis, organisms routinely identified by API 0E as Citrobacter species and Salmonella species. Of the tests used for identification by Sensititre, the API 0E lacks malonate, esculin, adonitol, and maltose but has instead arginine dihydrolase, glucose (included in Sensititre control row ), melibiose, and amygdalin. The species identification of Citrobacter by Sensititre is facilitated Downloaded from on December, 08 by guest

6 65 STANECK ET AL. J. CLIN. MICROBIOL. TABLE. Errors of identification by API 0E or Sensititre Taxon by API Sensititre conventionalnoofo.f biochemicals Response error. Response enrorosf Enteric group 7 Enteric group 7 Y. enterocolitica E. cloacae E. agglomerans E. cloacce E. coli H. alvei NMTb E. coli Pasteurella multocida S. sonnei E. coli E. coli Shigella sp. K. ozaenae K. rhinoscleromatis K. ozaenae K. pneumoniae E. agglomerans NMT K. pneumoniae S. rubidaea K. pneumoniae K. pneumoniae E. agglomerans E. cloacae P. mirabilis P. mirabilis M. morganii S. choleraesuis No ID" Salmonella pullorum S. marcescens S. marcescens S. liquefaciens S. marcescens S. liquefaciens S. marcescens S. liquefaciens S. marcescens 5 S. liquefaciens S. liquefaciens S. marcescens NMT a Valid identification response by API or Sensititre differing from taxon defined by conventional biochemical testing. b NMT, No matching taxa. ' Code number not in API data base. by the malonate and adonitol. API 0E has chosen not to include a species designation for most strains of Salmonella, because of either test limitations or corporate philosophy and instead strongly recommends, as does Sensititre, serological verification of both Salmonella and Shigella species. Correct identification at the genus level was made for approximately 97 and 98% of the isolates, respectively, by Sensititre and API 0E. For purposes of this study, numerical codes with low selectivity responses at the genus level were grouped along with code numbers not appearing in the data base as no ID responses. It could certainly be argued that the low selectivity response is useful in that certain taxonomic choices are given in order of likelihood probability, and the user is therefore free to either accept the most probable choice or perform additional tests to distinguish between the choices. Among the 5 no ID responses of API 0E listed in Table, only 7 were actually not listed in the data base. Of the 45 Sensititre no ID responses, however, 30 responses were not found in the data base. This observation could possibly be attributed to the relative maturity of the API 0E data base versus the Sensititre. The younger Sensititre data base would be expected to have a greater number of nonmatching codes. It is interesting to note that except for a cluster of six instances when the API system identified S. liquefaciens as S. marcescens, the number of incorrect responses (Table ) from either system was similar and represented a very small portion of the total number of isolates tested. The most serious errors, considering the clinical implications, were the three instances in which E. coli was called Shigella by Sensititre. However, in each of these instances the computer display instructed the user to confirm this identification by serology, and had the serology component of the identification procedure been carried out, as is the custom in most laboratories, an erroneous report would not have been made. Aside from these, the majority of responses other than complete agreement found in either system were scattered among the taxa studied. Although the present study was confined to identification of Enterobacteriaceae, the Sensititre data base also contains codes for various glucose-nonfermenting organisms. Casual observations throughout the study suggested that Pseudomonas aeruginosa is readily identified by Sensititre; however, the performance of Sensititre for other glucose-nonfermenting organisms awaits evaluation. The API 0E has addressed the non-glucose-fermenting group by providing an option for additional tests, increased incubation times when necessary, and expansion of the octal code to nine digits. The Sensititre system was reasonably simple and easy to use. The dried test reagents allow for refrigerator storage of the plates and long shelf life. The automated inoculator was efficient and rapid in the delivery of inocula to the plate. Plastic templates supplied with the system simplified and reduced errors in oil overlay, cover perforation, and later reagent addition to the plates. The test reactions were in general easily read with little practice, and the indicator lights of the plate viewer box greatly reduced the occasional confusion one encounters in reading Downloaded from on December, 08 by guest

7 VOL. 7, 983 a microtiter plate containing 96 wells of identical size and shape. During the course of the study, certain observations were made regarding the reading procedure, which facilitated the recognition of true test results. As with any new test system, strict guidelines needed to be followed in the reading of reactions to maximize performance and reproducibility. It was gratifying to note the small amount of variation in the reading of reactions between laboratories with Sensititre as well as API 0E described in the interlaboratory reproducibility studies, which attests to the importance of adherence to the reading guidelines. A white translucent Plexiglas sheet suspended above the Sensititre plate acted as a light diffuser and provided a white, evenly illuminated background against which the Sensititre test colors could be read, greatly improving the ease and consistency of reaction reading. The recommendation has been made to the manufacturer to incorporate this feature into the plate viewing box. It was also noted that four rather than two drops of Kovacs reagent added to the Sensititre indole well allowed the development color to spread throughout the entire well. With only two drops, a positive reaction would appear as a red ring, often difficult to detect when using the plate viewer. The Sensititre oxidase test was the only reaction that required reading immediately after reagent addition. Procedurally, therefore, all reagents except the oxidase reagent were added, and sufficient time was allowed for the slower reactions to develop. The oxidase reagent was then added immediately before plate reading. Mastery of the microcomputer component of Sensititre was easy due to a software program designed to be "user friendly." Simple instructions given in sequence by the program allow the user to initiate and complete the identification process in straightforward fashion. Although the edition of the user program employed in this study allowed for test alteration only after the completion of all reaction entries for a given organism, a more recent program allows the user to change a reaction immediately should a keyboard entry or reading error occur. Data interpretation could either be displayed or printed after entry of each organism's test results or be stored and printed after the reading of all plates. The latter option allows for rapid, uninterrupted reading and lends itself well to use of the system in batch fashion. It should be mentioned that although the multicenter study utilized the automated inoculator and data interpretation system, the Sensititre identification system can be utilized in a totally manual fashion by employing an eight-channel SENSITITRE BACTERIAL IDENTIFICATION SYSTEM 653 repeating pipette inoculator, a simple mirrored plate viewer, and a printed bench code booklet. Used as a manual system, the Sensititre offers considerable economic advantage over several commercially available multitest kits, including API 0E. The economic impact of use of the automated system is dependent on user test volume and the specific purchase arrangements offered by the vendor. The microplate itself, with a capacity for identification offour organisms, offers potential savings on consumable expenditures. Labor savings are realized via the automatic inoculation delivery versus manual inoculation, and the computerized data interpretation takes less time than the use of printed data code books. The time required for inoculum preparation, reagent addition, and reaction reading is comparable to manual multitest systems. Among the distinct advantages of the Sensititre equipment is its flexibility of use. Both the inoculator and the computer-assisted data interpretation module can be used with the Sensititre antimicrobial susceptibility system, which features various data management programs for test results. The Sensititre system also presents to microbiology laboratories the advantage of a fully operational microcomputer in the Apple II component of the system. The microcomputer, utilizing floppy disks, allows for easy program update and modifications of the Sensititre system. The usefulness of the computer is not limited to Sensititre programs in that it allows for the use of nonmicrobiological, statistical, educational, and data management programs. In addition, the option exists for user programming specific to the microbiological, epidemiological, or administrative needs of the laboratory. Among the possible limitations of the Sensititre Enterobacteriaceae identification system is the need for 8 to 4 h of incubation time. This would be problematic to laboratories desiring a rapid or same-day identification. Several identification kits presently offer 4- to 5-h identifications, including API 0E which offers a 5-h data profile index. In addition, the Sensititre program as it currently exists provides no further suggestions regarding biochemical tests needed to resolve low selectivity choices, a feature provided by the API 0E index. In summary, the Sensititre system appears to be a reliable and convenient method for the identification of Enterobacteriaceae. Sensititre can be utilized either as a manual system providing distinct economic advantages or as a semiautomated microcomputer-assisted system offering economical running costs and the flexibility of use with microtiter antimicrobial susceptibility testing plus commercially available or userwritten data management programs. Downloaded from on December, 08 by guest

8 654 STANECK ET AL. ACKNOWLEDGMENTS We thank Arthur Starr of Tampa, Fla., for his interest and support throughout this study. The technical expertise and assistance given by Marie Gourdeau and Marielle Parent of Montreal and Marlene Willner, Sherene Mackos, Barbara Puthoff, William Bayer, and Barbara Laymon of Cincinnati are gratefully acknowledged. We also thank Sharon Vincent and Steven Glenn of Cincinnati for their technical and editorial contributions and Joyce Turner of Cincinnati for her secretarial assistance in the preparation of this manuscript. LITERATURE CITED. Barry, A. L., R. E. Badal, and L. J. Effinger Identification of Enterobacteriaceae in frozen microdilution trays prepared by Micro-Media Systems. J. Clin. Microbiol. 0: Brenner, D. J., J. J. Farmer, F. W. Hickman, M. A. Asbury, and A. G. Steigerwalt Taxonomic and nomenclature changes in Enterobacteriaceae. Department of Health, Education and Welfare publication no. (CDC)78- J. CLIN. MICROBIOL Centers for Disease Control, Atlanta, Georgia. 3. Edwards, P. R., and W. H. Ewing (ed.). 97. Identification of Enterobacteriaceae. 3rd ed. Burgess Publishing Co., Minneapolis, Minnesota. 4. Kelly, S. A., and J. A. Washington II Evaluation of Micro-Media Quad Panels for identification of the Enterobacteriaceae. J. Clin. Microbiol. 0: Skerman, V. B. D., V. McGowan, and P. H. A. Sneath Approved lists of bacterial names. Int. J. Syst. Bacteriol. 30: Smith, P. B., K. M. Tomfohrde, D. L. Rhoden, and A. Balows. 97. API system: a multitube micromethod for identification of Enterobacteriaceae. Appi. Microbiol. 4: Thrupp, L. D Susceptibility testing of antibiotics in liquid media, p In V. Lorian (ed.), Antibiotics in laboratory medicine. The Williams & Wilkins Co., Baltimore. 8. Wilcox, W. R., S. P. Lapage, and B. Holmes A review of numerical methods in bacterial identification. Antonie van Leeuwenhoek J. Microbiol. Serol. 46: Downloaded from on December, 08 by guest