Evaluation of the BBL Crystal Anaerobe Identification System

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, Dec. 1997, p Vol. 35, No /97/$ Copyright 1997, American Society for Microbiology Evaluation of the BBL Crystal Anaerobe Identification System JOSEPH J. CAVALLARO,* LOIS S. WIGGS, AND J. MICHAEL MILLER Diagnostic Microbiology Section, Hospital Infections Program, Centers for Disease Control and Prevention, Atlanta, Georgia Received 19 May 1997/Returned for modification 7 July 1997/Accepted 2 September 1997 The BBL Crystal Anaerobe (ANR) identification system was evaluated, and the results were compared with those from conventional anaerobic methods. We tested 322 clinically significant anaerobic bacteria according to the manufacturer s instructions. The system identified correctly 286 of 322 (88.8%) of the anaerobic bacteria tested. Of these, 263 of 322 (81.7%) were identified correctly on initial testing and 49 were identified correctly only to the genus level; on repeat testing, 23 of 49 (46.9%) were identified correctly to both the genus and the species levels. A total of 26 (8.5%) were misidentified at the species level, and 10 (3.1%) were not identified. Performance characteristics for individual strains varied. The system correctly identified all tested strains of Campylobacter, Desulfomonas, Desulfovibrio, Leptotrichia, Mobiluncus, Peptostreptococcus, Porphyromonas, Provetella, Propionibacterium, Tisierella, and Veillonella and 36 of 37 (97.3%) Actinomyces strains, 42 of 46 (91.3%) B. fragilis group strains, 79 of 103 (76.7%) Clostridium strains, (however, the system failed to identify any of the 7 C. innocuum and 9 C. tetani strains tested), and 8 of 15 (53.3%) Bacteroides strains. This system was easy to use, did not involve the addition of reagents, and was faster than conventional anaerobic procedures. It would be a useful addition to the anaerobe laboratory of most hospitals. Clinical specimens submitted for anaerobic culture frequently contain complex mixtures of both aerobic and anaerobic organisms. Conventional methods for isolating and identifying anaerobes involve culture, time-consuming biochemical testing under strict anaerobic conditions, and gas-liquid chromatography before a definitive identification can be made (1, 2). These gold standard procedures are costly and often are not practical for many clinical microbiology laboratories. One means of reducing costs is by using simple and rapid tests for the identification of anaerobes. In recent years, several rapid-identification systems for anaerobic bacteria have been developed and evaluated, such as RapID ANA II (Innovative Diagnostic Systems, Inc., Atlanta, Ga.) (3 6, 16), ANI Card (Vitek Systems, Hazelwood, Mo.) (19), and AN-IDent and API ZYM (Analytab, Plainview, N.Y.) (3, 4, 15, 18 20). Recently, Becton Dickinson introduced the BBL Crystal Anaerobe (ANR) Identification (ID) system, which is a miniaturized identification method employing modified conventional, fluorogenic, and chromogenic substrates (12, 14) to identify anaerobic bacteria frequently isolated from clinical specimens without the need for anaerobic incubation. In this study, we evaluated the Crystal ANR ID system for its ability to identify clinical anaerobe isolates previously identified by the anaerobe reference laboratory at the Centers for Disease Control and Prevention (CDC). (This study was presented in part at the 97th General Meeting of the American Society for Microbiology, Miami Beach, Fla., 4 to 8 May 1997.) * Corresponding author. Mailing address: Centers for Disease Control and Prevention, 1600 Clifton Rd., NE, Mail Stop C-16, Atlanta, GA Phone: (404) Fax: (404) MATERIALS AND METHODS Bacterial strains. To establish performance characteristics, we evaluated 307 anaerobes. All bacteria tested were either clinical isolates obtained from specimens submitted to the CDC anaerobe reference laboratory or strains from the American Type Culture Collection (ATCC). These organisms had been identified to genus and species levels by conventional reference identification methods and gas-liquid chromatography (8 10, 12, 13, 21). The test organisms were stored lyophilized at 4 C. All lyophilized strains were subcultured at least twice on CDC anaerobe blood agar plates (Carr-Scarborough Microbiologicals, Stone Mountain, Ga.) before being used in the study. As recommended by the kit manufacturer, Bacteroides fragilis ATCC was used for the quality control testing of each new lot of test panels. BBL Crystal ANR ID kit. The BBL Crystal ANR ID kit consists of 20 panel lids, 20 bases, and 20 tubes of inoculum fluid. Many of the tests used in this system are modifications of classical methods and include tests for fermentation, oxidation, and degradation or hydrolysis of various substrates. In addition, this system uses chromogenic and fluorochrome-linked substrates to detect preformed metabolic enzymes. The panel lids contain 30 plastic prongs; the tips of 29 of the prongs each contain a different enzymatic substrate, and 1 contains a fluorescence control. The bases have a total of 30 reaction wells. A bacterial suspension prepared in the inoculum fluid is used to fill all 30 wells. When the lid is aligned with the base and snapped in place, the bacterial inoculum rehydrates the dried substrates on the prong tips and initiates the test reactions. Test procedure. The Crystal ANR ID system is not intended for use directly with clinical specimens. Therefore, pure cultures of isolates must be prepared from a nonselective blood agar medium for use as the inoculum. In addition, the Crystal ANR ID system requires Gram stain, catalase (3% H 2 O 2 ), and indole (spot test with 1% p-dimethylaminocinnamaldehyde) results for each isolate tested. Lyophilized strains were reconstituted and were passed at least twice on CDC anaerobe blood agar in an anaerobic environment at 35 C. After 24 to 48 h of incubation (for as much as 72 h for some slowly growing cocci and Actinomyces species), colonies from each strain were harvested aseptically with cotton-tipped applicator swabs and suspended in a labeled tube of inoculum fluid to a turbidity equivalent to a McFarland no. 4 standard (not to exceed a McFarland no. 5 standard). Each inoculum tube was vortexed for 10 to 15 s, and the entire contents were poured into an appropriately labeled panel base. The inoculum was then gently rolled along the tracks of the base until all of the reaction wells were filled. A lid was aligned over each base and snapped into place. The inoculated panels were placed in incubation trays and incubated for 4 h in a non-co 2 incubator with 40 to 60% humidity at 35 C. After incubation, all panels were read visually with the BBL Crystal Panel Viewer and a 10-digit profile number was generated and recorded on a pad listing results. The profile number and off-line spot biochemical test results (catalase and indole) and Gram stain reaction were entered into a computer on which the BBL Crystal ID System Electronic Codebook had been installed. The appropriate database was selected based on the type of primary plate used to prepare the test inoculum (in this case, CDC anaerobe blood agar). The computer program generates a single genus and species identification or several possible identifications, along with biotype validity and confidence levels, comments, and any atypical test results. In cases in which the test results were not sufficiently discriminatory to provide a definitive identification, a differentiation database was included to assist in resolving pairs of organisms that occasionally could not be differentiated on the basis of the profile and off-line tests alone. A complete list of taxa that comprises 3186

2 VOL. 35, 1997 EVALUATION OF BBL CRYSTAL ANR ID 3187 FIG. 1. Taxa in the BBL Crystal ANR ID System. the BBL Crystal ANR ID database is shown in Fig. 1. Identification of test organisms was derived from a comparative analysis of the reaction patterns of the test isolates to those held in the database. For the purposes of this study, an identification was considered correct when the genus and species reported by the Crystal ANR ID system matched, on initial testing, the previously recorded conventional identification. When the initial test result generated a correct genus but an incorrect species, the test was repeated; when the repeated test was correct to both genus and species levels, the final test result was recorded as correct. When the species result was still incorrect, an error was recorded.

3 3188 CAVALLARO ET AL. J. CLIN. MICROBIOL. TABLE 1. Identification of anaerobic gram-negative bacilli with the BBL Crystal System Species No. correct/total (%) No. correct ID on retest Error No ID Total (% correct) B. fragilis group 34/46 (73.9) /46 (91.3) B. caccae B. distasonis B. fragilis B. levii B. ovatus B. splanchnicus B. thetaiotaomicron B. uniformans B. vulgatus Bacteroides spp. 8/15 (53.3) /15 (53.3) B. capillosus B. gracilis B. ureolyticus Bilophila 1/1 (100) /1 (100) B. wadsworthia Campylobacter 6/6 (100) /6 (100) C. rectus/c. curvus Desulfomonas 2/2 (100) /2 (100) D. pigra Desulfovibrio 3/3 (100) /3 (100) D. desulfuricans Fusobacterium 25/26 (96.2) 1 1* 0 26/26 (100) F. gonidiaformans a 0 0 1* 0 F. mortiferum F. necrophorum F. nucleatum F. russii F. varium Leptotrichia 7/7 (100) /7 (100) L. buccalis Porphyromonas 1/2 (50) /2 (100) P. asaccharolytica P. endodontalis Prevotella 11/12 (91.7) /12 (100) P. bivia P. denticola P. disiens P. intermedia Tissierella 1/1 (100) /1 (100) T. praecuta Total 99/121 (81.8) 11/12 (91.7) 8/121 (6.6) 3/121 (2.5) 110/121 (90.9) a Not included in data analysis. When the Crystal ANR ID system provided an incorrect genus, the test was repeated. When the repeat test was incorrect, the final result was recorded as an error. However, when the result of the repeat test was correct to both genus and species levels, the test was again repeated. When the second repeat test was correct, the final result was recorded as correct. The percent accuracy of the BBL Crystal ANR ID system was calculated as the final percent correct of the total number of strains tested. The percent error of the system was calculated as the final percent incorrect. RESULTS The BBL Crystal ANR ID system correctly identified 286 of 322 (88.8%) of the strains tested. Of these, 263 of 322 (81.7%) were identified correctly on initial testing and 49 were identified correctly only to the genus level; on the repeat test, 23 of 49 (46.9%) were identified correctly to both genus and species levels. A total of 26 of 307 strains (8.5%) were misidentified, and 10 of 322 (3.1%) resulted in no identification. Performance characteristics for individual strains varied. Table 1 shows the results for the anaerobic gram-negative bacilli. Of the 46 B. fragilis group strains tested, 42 (91.3%) were identified correctly. A total of 9 of 12 B. fragilis strains (75%) were identified correctly on initial testing, and 3 of 12 (25%) were identified correctly on repeat testing. Four of seven Bacteroides thetaiotamicron strains were identified as B. fragilis initially, and as either Bacteroides stercoris or Bacteroides uniformis on retesting; conventional anaerobic testing confirmed these isolates as being B. thetaiotamicron. One of the Bacteroides distasonis strains was misidentified initially as Eubacterium aerofaciens, and another was reported as no identification; upon retesting, both were identified correctly as B. distasonis. None of the seven Bacteroides gracilis strains tested were correctly identified; three were reported as no identification, three were reported as Bacteroides ureolyticus, and one was reported as Tissierella praeacuta. Repeat testing did not resolve the errors, but conventional anaerobic testing confirmed that these isolates were B. gracilis. Two isolates of Desulfomonas pigra and two isolates of Desulfovibrio desulfuricans were identified as gram-negative rods that were indole negative and catalase positive. The electronic database lists three possible species in this category, two of which are Desulfovibrio species and D. pigra; repeated testing produced the same results. One D. desulfuricans isolate was identified as a gram-negative rod that was indole negative and

4 VOL. 35, 1997 EVALUATION OF BBL CRYSTAL ANR ID 3189 TABLE 2. Identification of anaerobic gram-positive bacilli with the BBL Crystal System Species No. correct/total (%) No. correct ID on retest Error Total (% correct) Actinomyces 31/37 (83.8) /37 (97.3) A. israelii A. odontolyticus A. meyeri A. naeslundii A. pyogenes A. viscosus Bifidobacterium 2/3 (66.7) 1 0 3/3 (100) B. adolescentis B. breve Clostridium 76/103 (73.8) /103 (76.7) C. baratii C. beijerinckii C. butyricum C. clostridioforme C. difficile C. innocuum C. perfingens C. ramosum C. septicum C. sordelii C. sporogenes C. tertium C. tetani Eubacterium 5/6 (83.3) 1 0 6/6 (100) E. limosum E. lentum Mobiluncus 7/7 (100) 0 0 7/7 (100) M. curtisii M. mulieris Propionibacterium 26/26 (100) /26 (100) P. acnes P. avidum P. granulosum Total 147/182 (80.8) 10/167 (5.9) 25/182 (13.7) 157/182 (86.3) catalase negative. The electronic database lists three possible species included in this category, one of which is a Desulfovibrio species; repeat testing produced the same result. Since the correct identification was among the three choices generated by the computer program, these results were recorded as correct. Of the Fusobacterium strains tested, 25 of 26 (92.6%) were identified correctly on initial testing. A single isolate of Fusobacterium gonidiaformans, which is not included in the database, was misidentified twice as either Fusobacterium necrophorum or Fusobacterium varium. This was the only Fusobacterium strain that was not in the CDC anaerobe blood agar database and, as such, was not included in the final statistical evaluation. One F. necrophorum isolate was reported to be F. varium/f. necrophorum on initial testing; on retesting it was identified correctly as F. necrophorum. All seven Leptotrichia buccalis isolates were identified correctly on initial testing. One Porphyromonas asaccharolytica isolate was misidentified as B. thetaiotaomicron/b. uniformans on initial testing but was correctly identified on repeat testing. A total of 11 (91.7%) of 12 Prevotella strains were identified correctly on initial testing. Although one Prevotella bivia isolate was identified as P. bivia/prevotella melaninogenica and one Prevotella intermedia isolate was identified as Prevotella disiens/p. intermedia, both isolates were identified correctly on repeat testing. One Tisierella praecuta strain was identified correctly. In this study, the Crystal ANR ID system correctly identified 110 of 121 (90.9%) gram-negative bacilli strains tested; 3 of 121 (2.5%) were reported as No Identification, and 8 of 121 (6.6%) were misidentified. Table 2 presents results for the anaerobic gram-positive bacilli. A total of 37 Actinomyces isolates or strains were tested; of these, 36 (97.3%) were identified correctly. Two Actinomyces naeslundii strains were misidentified as Bifidobacterium dentium on initial testing; on repeat testing, one was correctly identified as A. naeslundii, and the other was again misidentified as B. dentium/actinomyces israelii. Both isolates were identified as A. naeslundii by conventional anaerobic methods. Of the 103 Clostridium isolates tested, 79 (76.7%) were identified correctly; all strains of C. beijerinckii, C. butyricum, C. clostridiforme, C. perfringens, C. ramosum, C. sordellii, and C. tertium were identified correctly on initial testing. Six of nine C. tetani isolates were misidentified on initial testing as C. bifermentans; for the remaining C. tetani isolates, the Crystal ID system reported the following choices: for two, either nonreactive gram-positive rods which are indole positive or C. bifermentans, and for one, either nonreactive gram-positive rods which are indole positive or C. cadaveris. In both instances, the Crystal ID system recommended supplemental testing as it was not able to definitively distinguish between the two possible choices. Repeat testing produced the same results. The Crystal ANR ID system also failed to identify seven C. innocuum isolates and misidentified seven C. sporogenes isolates; six C. sporogenes isolates were misidentified as C. sporogenes/c. subterminale, and one was misidentified as E. aerofaciens. Two retests yielded the same results.

5 3190 CAVALLARO ET AL. J. CLIN. MICROBIOL. TABLE 3. Identification of anaerobic gram-positive and gram-negative cocci with the BBL Crystal System Species No. correct/ total (%) Among the anaerobic non-spore-forming gram-positive bacilli tested, two Bifidobacterium adolescentis isolates were correctly identified. One Bifidobacterium breve isolate was initially misidentified as B. adolescentis but was identified as B. breve on two retests. Four of five Eubacterium lentum strains were identified correctly, and one was initially misidentified as E. aerofaciens but was identified as E. lentum on two retests. All of the Mobiluncus and Propionibacterium species tested were identified correctly. In summary, the Crystal ANR ID system correctly identified 157 of 182 (86.3%) of the anaerobic grampositive bacilli tested, while 18 of 167 (10.8%) were misidentified and 7 of 182 (3.8%) were reported as no identification. Table 3 summarizes results for the anaerobic cocci. A total of 15 of 17 (88.2%) Peptostreptococcus isolates and 2 of 2 (100%) Veillonella parvula strains were identified correctly on initial testing. Of two Peptostreptococcus prevotii isolates tested, one was misidentified as Peptostreptococcus anaerobius, and one was reported as no identification on initial testing. On repeat testing, both isolates were correctly identified. DISCUSSION No. correct ID on retest Error Total (% correct) Peptostreptococcus 15/17 (88.2) /17 (100) P. anaerobius P. asaccharolyticus P. indolicus P. magnus/p. micros P. prevotii P. tetradius Veillonella 2/2 (100) 0 0 2/2 (100) V. parvula Total 17/19 (89.5) 2/19 (10.5) 0 19/19 (100) In this study, the BBL Crystal ANR ID system was evaluated for its ability to identify a wide variety of anaerobic bacteria. After an initial period of familiarization with the color reactions with the regular (white) light source and the UV light source (for fluorescent substrates), the reading of the panels and interpretation of test results were not difficult. Enzymatic hydrolysis of fluorogenic substrates resulted in an increase in the level of fluorescence, compared to that of the negative control well, that was easily detected visually with a UV light source. Upon hydrolysis, chromogenic substrates produced a gold-yellow color that was easily detected visually with a white light source; negative hydrolysis produced a colorless result. On the worksheet, each positive test result was given a value of 4, 2, or 1; a value of 0 was given to any negative result. The values resulting from each positive reaction in each column of the worksheet were added together, and a 10-digit profile number was generated. All data in this study were based on results obtained with the BBL Crystal ID Systems software version 3.0 (specifically, BBL Anaerobe from CDC Blood Agar 3.0). It is generally recognized that many different species of anaerobes can potentially be isolated from clinical specimens; however, the actual number of species routinely isolated is relatively small. It has been reported by Finegold (11) that only five anaerobes or groups of anaerobes account for two-thirds of clinically significant, anaerobe-associated infectious processes. These are anaerobic cocci, members of the B. fragilis group, pigmented Bacteroides species (recently reclassified as Porphyromonas and Prevotella spp.), Fusobacterium nucleatum, and Clostridium perfringens. In our study, the Crystal system performed well in the identification of these critical anaerobes and a variety of other significant anaerobic bacteria. As previously noted, the Crystal system correctly identified 100% of the Peptostreptococcus and Veillonella strains tested. Likewise, 100% of the Porphyromonas and Prevotella strains and F. nucleatum tested were identified correctly. Members of the B. fragilis group are the most commonly recovered anaerobes found in clinical specimens. In general, it is important to identify members to the species level since members of this group are more resistant to antimicrobial agents than most other anaerobes and exhibit species-to-species variability in both virulence and drug resistance. In this study, the Crystal system identified to the species level, on initial testing, all of the following members of the B. fragilis group tested: B. caccae, B. levii, B. splanchnicus, and B. uniformans. While the system did correctly identify B. fragilis strains on initial testing and repeat testing, it did have difficulty in correctly identifying the B. thetaiotaomicron strains tested. Four of seven strains were initially identified as B. fragilis by the system. As a rule, B. fragilis tends to be significantly more susceptible to certain antimicrobial agents (cephalosporins and clindamycin) than B. thetaiotaomicron, B. vulgatus, and B. ovatus. Therefore, this difficulty of the Crystal system in correctly identifying the B. thetaiotaomicron strains tested should be of concern to clinical laboratories not performing susceptibility testing on anaerobes, since an incorrect Crystal system identification for this organism may subsequently result in inappropriate antibiotic therapy. Similarly, difficulty was experienced in identifying B. gracilis, a member of the B. urolyticus group. All seven of the B. gracilis strains tested were identified incorrectly; three of these were identified as B. urolyticus. Species identification within this group is important because B. gracilis is known to be more pathogenic than B. urolyticus and is often resistant to antimicrobial agents, including penicillins and cephalosporins. The clostridia presented the BBL Crystal system with its greatest challenge. The system correctly identified only 79 of 103 (76.7%) Clostridium strains tested. The result is skewed due to the inability of the system to identify C. innocuum (0 of 7), C. sporogenes (0 of 7), and C. tetani (0 of 9). The Crystal system misidentified six of the seven C. sporogenes isolates as C. sporogenes/c. subterminale, indicating that the system was unable to distinguish between the two species; two repeat tests yielded the same results. One of the tested C. sporogenes isolates was identified as a E. aerofaciens. Two repeat tests yielded the same results. The Crystal ANR ID system is known to yield a small percentage of results such that there is a high probability that an isolate may be one of two (or less commonly, three) possible species. These instances usually represent species from the same genus or species that are closely related in their metabolic activities. This occurred with C. sporogenes. The Crystal result alone was inadequate to distinguish among the reported organisms. However, as recommended, these two species were differentiated successfully by using supplemental tests. The system also misidentified nine of nine C. tetani isolates; eight isolates were identified as C. bifermentans and one was identified as C. cadaveris in three consecutive tests. This is a major shortcoming, since the ability to expeditiously identify C. tetani would be of great importance to the clinician in initiating appropriate therapy. Any competent laboratory engaged in anaerobic cultivation should be able to isolate in pure culture any Clostridia present in a clinical specimen. For example, in this circumstance, when preliminary evidence (Gram

6 VOL. 35, 1997 EVALUATION OF BBL CRYSTAL ANR ID 3191 stain and anaerobic colonial morphology and characteristics, etc.) indicates the presence of C. tetani (strict anaerobe, swarming colonies, and drumstick-shaped bacillus with terminal spores), appropriate supplemental tests would be warranted to differentiate successfully C. tetani, C. cadaveris, and C. bifermentans (10, 12, 21). It is important to note that the Crystal system did perform well in correctly identifying two of three Clostridium isolates that are difficult to stain with Gram stain and that routinely appear gram variable or gram negative. Thus, eight of eight C. clostridiforme and seven of seven C. ramosum isolates were correctly identified. Although we interpreted and recorded the Gram stain preparations for these organisms as gram positive, we challenged the system by also coding duplicate entries as gram negative. In all cases, whether the entries were coded as gram positive or gram negative, a correct identification was made by the system. This was not unexpected, since identification results for the BBL Crystal ID systems are derived from a probability matrix of 30 results and recommended off-line tests (i.e., indole, catalase, and Gram stain). Thus, an unusual off-line test is treated by the system in the same manner as that for any test result, and a profile with unexpected off-line results may still yield an acceptable identification when most other Crystal results agree with the expected database values. The seven strains of C. innocuum tested were reported initially and upon retesting as no identification. Since this organism may stain in a gram-variable manner, it was coded both as gram positive and gram negative; in both cases, they were reported as no identification. Conventional anaerobic testing confirmed these isolates as being C. innocuum. The experience with the Clostridium isolates showed that the system is not without limitations. Since the Crystal ANR ID system uses a modified microenvironment, the results for its biochemical tests may differ from those previously established with conventional biochemical tests, thus making direct comparison difficult. In addition, it is generally recognized that minor variations may exist in strains within species. The isolates that we tested included a mixture of commonly isolated anaerobic species and others that had been phenotypically unusual and/or difficult to identify; this provided a challenge to the sensitivity of the system. Thus, the performance of the system in this study may be a reflection of both the difficulty in working with the latter group of organisms and the limitations of the system s database. For example, when a test profile yielded a no identification result and the culture purity had been confirmed, then one would need to consider that the tested species or the specific profile number may not have been included in the database. This was shown by the inability of the system to recognize a single isolate of F. gonidioformans. Although in this study we chose to use CDC anaerobe blood agar as the nonselective primary medium, the system can be used with isolates from any one of four different nonselective media (CDC anaerobe blood agar, brucella blood agar, Columbia blood agar, or Schaedler agar with vitamin K 1 and 5% sheep blood). However, previously published reports indicate that the results may differ when isolates from one of the other nonselective media are used (7, 17). Users of this system should recognize the limitations discussed above. As with all anaerobic bacteriology work, the interpretation of results generated by this system requires the application of basic microbiological knowledge and recognition of certain key characteristics of anaerobic bacteria encountered in the clinical microbiology laboratory. When warranted, the final identification of isolates should take into consideration the possible need for additional conventional anaerobic testing. In this respect, we recommend as excellent reference sources the texts by Engelkirk et al. (10) and Summanen et al. (21) and the laboratory manual by Holdeman et al. (12). In our study, the BBL Crystal ANR ID system was shown to be an acceptable 4-h system for the identification of most clinically significant anaerobic bacteria. The list price for the BBL Crystal ANR ID system is $139 per 20 test kits, or $6.95 per test. This per-test cost is similar to those for the RapID ANA II and An-IDENT tests (18). The system was easy to use, required no additional reagent test-developing steps, and was faster than conventional techniques. This system would be a useful addition to the anaerobe laboratories of most hospitals. REFERENCES 1. Ballows, A., W. J. Hausler, Jr., K. L. Herrmann, H. D. Isenberg, and H. J. Shadomy (ed) Manual of clinical microbiology, 5th ed. American Society for Microbiology, Washington, D.C. 2. Baron, E. J., and S. M. Finegold Bailey and Scott s diagnostic microbiology, 8th ed. C. V. Mosby Company, St. Louis, Mo. 3. Bate, G Comparison of Minitek Anaerobe II, API AnIdent, and RapID ANA systems for identification of Clostridium difficile. Am. J. Clin. Pathol. 85: Brander, M. A., and H. R. Jousimies-Somer Evaluation of the RapID ANA II and API ZYM systems for identification of Actinomyces species from clinical specimens. J. Clin. Microbiol. 30: Burlage, R. S., and P. D. Ellner Comparison of the PRAS II, AN- Ident, and RapID-ANA systems for the identification of anaerobic bacteria. J. Clin. Microbiol. 22: Celig, D. M., and P. C. Schreckenberger Clinical evaluation of the RapID-ANA II panel for the identification of anaerobic bacteria. J. Clin. Microbiol. 29: Daly, J. A., J. E. Castro, and E. K. Korgenski Evaluation of the BBL Crystal Anaerobe System for the identification of clinical anaerobic isolates, abstr. C-359, p. 64. In Abstracts of the 95th General Meeting of the American Society for Microbiology. American Society for Microbiology, Washington, D.C. 8. Dowell, V. R., Jr., G. L. Lombard, F. S. Thompson, and A. Y. Armfield Media for isolation, characterization, and identification of obligately anaerobic bacteria. U.S. Department of Health and Human Services, Centers for Disease Control, Atlanta, Ga. 9. Dowell, V. R., Jr., and T. M. Hawkins Laboratory methods in anaerobic bacteriology. U.S. Department of Health and Human Services, Centers for Disease Control, Atlanta, Ga. 10. Engelkirk, P. G., J. Duben-Engelkirk, and V. R. Dowell, Jr. (ed) Principles and practice of clinical anaerobic bacteriology. Star Publishing Company, Belmont, Calif. 11. Finegold, S. M Anaerobic bacteria: their role in infection and their management. Postgrad. Med. 81: Holdeman, L. V., E. P. Cato, and W. E. C. Moore (ed) Anaerobe laboratory manual, 4th ed. Virginia Polytechnic Institute and State University, Blacksburg, Va. 13. Lombard, G. L., and V. R. Dowell, Jr Gas-liquid chromatography analysis of the acid products of bacteria. U.S. Department of Health and Human Services, Centers for Disease Control, Atlanta, Ga. 14. Manafi, M., W. Kneifel, and S. Bascomb Fluorogenic and chromogenic substrates used in bacterial diagnostics. Microbiol. Rev. 55: Marler, L., S. Allen, and J. Siders Rapid enzymatic characterization of clinically encountered anaerobic bacteria with the API ZYM system. Eur. J. Clin. Microbiol. 3: Marler, L. M., J. A. Siders, L. C. Wolters, Y. Pettigrew, B. L. Skitt, and S. D. Allen Evaluation of the new RapID-ANA system for the identification of clinical anaerobic isolates. J. Clin. Microbiol. 29: McTeague, M., P. Summanen, M. L. Vaisanen, and S. M. Finegold Evaluation of BBL crystal anaerobe identification panel for the identification of anaerobic bacteria, abstr. C-368, p. 64. In Abstracts of the 95th General Meeting of the American Society for Microbiology. American Society for Microbiology, Washington, D.C. 18. Miller, P. H., L. S. Wiggs, and J. M. Miller Evaluation of API An-IDENT and RapID ANA II systems for identification of Actinomyces species from clinical specimens. J. Clin. Microbiol. 33: Schreckenberger, P. C., D. M. Celig, and W. M. Janda Clinical evaluation of the Vitek ANI Card for identification of anaerobic bacteria. J. Clin. Microbiol. 26: Stenson, M. J., D. T. Lee, J. E. Rosenblatt, and J. M. Contezac Evaluation of the AnIDENT system for identification of anaerobic bacteria. Diagn. Microbiol. Infect. Dis. 5: Summanen, P., E. J. Baron, D. M. Citron, C. Strong, H. M. Wexler, and S. M. Finegold (ed) Wadsworth anaerobic bacteriology manual. Star Publishing Company, Belmont, Calif.